{"id":654,"date":"2026-11-20T18:05:54","date_gmt":"2026-11-20T18:05:54","guid":{"rendered":"https:\/\/blockoala.com\/index.php\/2026\/11\/20\/lars-reduces-memory-use-when-fine-tuning-ai-models-on-small-devices\/"},"modified":"2026-11-20T18:05:54","modified_gmt":"2026-11-20T18:05:54","slug":"lars-reduces-memory-use-when-fine-tuning-ai-models-on-small-devices","status":"publish","type":"post","link":"https:\/\/blockoala.com\/index.php\/2026\/11\/20\/lars-reduces-memory-use-when-fine-tuning-ai-models-on-small-devices\/","title":{"rendered":"LARS reduces memory use when fine-tuning AI models on small devices"},"content":{"rendered":"<p>Researchers at MIT\u2019s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Harvard University have developed a method to reduce the memory needed to fine-tune large language models. In reporting published on November 19, 2026, MIT CSAIL said the method, called Low-memory Activation Rank Subspace (LARS), addresses a memory cost that remains even when only a small fraction of a model\u2019s parameters are updated.<\/p>\n<p>Parameter-efficient fine-tuning methods such as LoRA limit parameter updates, but training still stores intermediate calculations known as <em>activations<\/em>. LARS reduces that storage burden by keeping those calculations in a more compact form. In the researchers\u2019 tests, it reduced memory use by more than 23% on GPUs and about 47% on CPUs compared with leading parameter-efficient fine-tuning methods.<\/p>\n<p>On an 8 GB Raspberry Pi 5, LARS saved 9% more memory than LoRA when fine-tuning at a context length of 128,000; the researchers found that savings grew as context length increased. In other tests, LARS saved more memory than the comparison methods but was slightly less accurate than LoRA. Across question-answering tests on NVIDIA GPUs, it saved about 10 GB more than its leading baseline. In CPU tests using prompts of different lengths, the researchers also observed roughly 22% lower power and energy use than comparable methods.<\/p>\n<p>MIT CSAIL researcher Irene Tenison said the approach could help devices analyze longer stretches of data locally\u2014for example, heart-rate readings on a fitness watch. That is a potential application, not a reported deployment. The researchers presented the LARS paper at the Asia-Pacific Chapter of the Association for Computational Linguistics conference in November.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Researchers at MIT\u2019s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Harvard University have developed a method to [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":653,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-654","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/posts\/654","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/comments?post=654"}],"version-history":[{"count":0,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/posts\/654\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/media\/653"}],"wp:attachment":[{"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/media?parent=654"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/categories?post=654"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blockoala.com\/index.php\/wp-json\/wp\/v2\/tags?post=654"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}